Add basic String#localeCompare (#1008)

MaxGraey · dcodeIO · commit 454e9e688dcd · 2020-01-10T15:21:56.000+01:00
diff --git a/std/assembly/index.d.ts b/std/assembly/index.d.ts
@@ -1417,6 +1417,7 @@ declare class String {
   endsWith(other: string): bool;
   indexOf(other: string, fromIndex?: i32): i32;
   lastIndexOf(other: string, fromIndex?: i32): i32;
+  localeCompare(other: string): i32;
   includes(other: string): bool;
   startsWith(other: string): bool;
   substr(start: i32, length?: i32): string;
diff --git a/std/assembly/string.ts b/std/assembly/string.ts
@@ -108,8 +108,8 @@ import { idof } from "./builtins";
   @operator(">") private static __gt(left: String | null, right: String | null): bool {
     if (left === right || left === null || right === null) return false;
     var leftLength  = left.length;
+    if (!leftLength) return false;
     var rightLength = right.length;
-    if (!leftLength)  return false;
     if (!rightLength) return true;
     // @ts-ignore: string <-> String
     return compareImpl(left, 0, right, 0, min(leftLength, rightLength)) > 0;
@@ -121,10 +121,10 @@ import { idof } from "./builtins";
 
   @operator("<") private static __lt(left: String, right: String): bool {
     if (left === right || left === null || right === null) return false;
-    var leftLength  = left.length;
     var rightLength = right.length;
     if (!rightLength) return false;
-    if (!leftLength)  return true;
+    var leftLength  = left.length;
+    if (!leftLength) return true;
     // @ts-ignore: string <-> String
     return compareImpl(left, 0, right, 0, min(leftLength, rightLength)) < 0;
   }
@@ -163,6 +163,17 @@ import { idof } from "./builtins";
     return -1;
   }
 
+  // TODO: implement full locale comparison with locales and Collator options
+  localeCompare(other: String): i32 {
+    if (other === this) return 0; // compare pointers
+    var len: isize = this.length;
+    var otherLen: isize = other.length;
+    if (otherLen != len) return select(1, -1, len > otherLen);
+    if (!otherLen) return 0; // "" == ""
+    // @ts-ignore: string <-> String
+    return compareImpl(this, 0, other, 0, otherLen);
+  }
+
   startsWith(search: String, start: i32 = 0): bool {
     if (search === null) search = changetype<String>("null");
     var len = <isize>this.length;
diff --git a/std/assembly/util/string.ts b/std/assembly/util/string.ts
@@ -101,11 +101,13 @@ export function compareImpl(str1: string, index1: usize, str2: string, index2: u
 }
 
 export function isSpace(c: i32): bool {
-  if (c <= 0xFF) {
+  if (c < 0x1680) { // < <LS> (1)
     // <SP>, <TAB>, <LF>, <VT>, <FF>, <CR> and <NBSP>
-    return c == 0x20 || <u32>(c - 0x09) <= 0x0D - 0x09 || c == 0xA0;
+    // (c == 0x20 || c == 0xA0) was optimized to (c | 0x80) == 0xA0
+    // @ts-ignore: cast
+    return ((c | 0x80) == 0xA0) | (u32(c - 0x09) <= 0x0D - 0x09);
   }
-  if (<u32>(c - 0x2000) <= 0x200A - 0x2000) return true;
+  if (u32(c - 0x2000) <= 0x200A - 0x2000) return true;
   switch (c) {
     case 0x1680: // <LS> (1)
     case 0x2028: // <LS> (2)
diff --git a/tests/compiler/std/string-encoding.optimized.wat b/tests/compiler/std/string-encoding.optimized.wat
@@ -2722,7 +2722,7 @@
    if
     i32.const 0
     i32.const 544
-    i32.const 684
+    i32.const 695
     i32.const 8
     call $~lib/builtins/abort
     unreachable
@@ -2745,7 +2745,7 @@
    if
     i32.const 0
     i32.const 544
-    i32.const 688
+    i32.const 699
     i32.const 8
     call $~lib/builtins/abort
     unreachable
@@ -3062,7 +3062,7 @@
   if
    i32.const 0
    i32.const 544
-   i32.const 702
+   i32.const 713
    i32.const 6
    call $~lib/builtins/abort
    unreachable
diff --git a/tests/compiler/std/string-encoding.untouched.wat b/tests/compiler/std/string-encoding.untouched.wat
@@ -4453,7 +4453,7 @@
    if
     i32.const 0
     i32.const 544
-    i32.const 684
+    i32.const 695
     i32.const 8
     call $~lib/builtins/abort
     unreachable
@@ -4477,7 +4477,7 @@
    if
     i32.const 0
     i32.const 544
-    i32.const 688
+    i32.const 699
     i32.const 8
     call $~lib/builtins/abort
     unreachable
@@ -4833,7 +4833,7 @@
   if
    i32.const 0
    i32.const 544
-   i32.const 702
+   i32.const 713
    i32.const 6
    call $~lib/builtins/abort
    unreachable
diff --git a/tests/compiler/std/string.optimized.wat b/tests/compiler/std/string.optimized.wat
diff --git a/tests/compiler/std/string.ts b/tests/compiler/std/string.ts
@@ -78,6 +78,17 @@ assert(str.lastIndexOf(", I", -1) == -1);
 assert(str.lastIndexOf("i", 0) == -1);
 assert(str.lastIndexOf("hi", 0) == 0);
 
+assert("".localeCompare("") == 0);
+assert("a".localeCompare("") == 1);
+assert("".localeCompare("a") == -1);
+assert("null".localeCompare("null") == 0);
+assert("abc".localeCompare("abd") == -1);
+assert("abd".localeCompare("abc") == 1);
+assert("abcd".localeCompare("abc") == 1);
+assert("abc".localeCompare("abcd") == -1);
+assert("".localeCompare("   ") == -1);
+assert("\0".localeCompare("") == 1);
+
 assert("".trimStart() == "");
 assert("ab c".trimStart() == "ab c");
 assert(" \n\t\rabc \t\r ".trimStart() == "abc \t\r ");
diff --git a/tests/compiler/std/string.untouched.wat b/tests/compiler/std/string.untouched.wat
